Multiline entry results in multiple entries

[Mulugruntz]

Let's say I have the following changelog:

# Changelog
All notable changes to this project will be documented in this file.

The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

## [Unreleased]
### Changed
- Changed `f()` and `g()` into `h(b:bool)`.

  **The change is NOT backward compatible!!** 
  
  Please replace `.f()` -> `.h()`
  And replace `.g()` -> `.h(b=True)`

  Reason: There is a bug in `somelib` ...

  More: https://bugs.python.org/issueXXXXX
- Release note 2.

## [1.0.0] - 2017-04-10
### Deprecated
- Known issue 1 (1.0.0)
- Known issue 2 (1.0.0)

[Unreleased]: https://github.test_url/test_project/compare/v1.0.0...HEAD
[1.0.0]: https://github.test_url/test_project/releases/tag/v1.0.0

Rendered as so:

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog,
and this project adheres to Semantic Versioning.

Unreleased

Changed

• Changed f() and g() into h(b:bool).

  The change is NOT backward compatible!!

  Please replace .f() -> .h()
  And replace .g() -> .h(b=True)

  Reason: There is a bug in somelib ...

  More: https://bugs.python.org/issueXXXXX

• Release note 2.

1.0.0 - 2017-04-10

Deprecated

• Known issue 1 (1.0.0)
• Known issue 2 (1.0.0)

If I run:

import keepachangelog
print(keepachangelog.from_dict(keepachangelog.to_dict('CHANGELOG.md', show_unreleased=True)))

This is what I get:

# Changelog
All notable changes to this project will be documented in this file.

The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

## [Unreleased]
### Changed
- Changed `f()` and `g()` into `h(b:bool)`.
- The change is NOT backward compatible!!**
- Please replace `.f()` -> `.h()`
- And replace `.g()` -> `.h(b=True)`
- Reason: There is a bug in `somelib` ...
- More: https://bugs.python.org/issueXXXXX
- Release note 2.

## [1.0.0] - 2017-04-10
### Deprecated
- Known issue 1 (1.0.0)
- Known issue 2 (1.0.0)

[Unreleased]: https://github.test_url/test_project/compare/v1.0.0...HEAD
[1.0.0]: https://github.test_url/test_project/releases/tag/v1.0.0

Rendered as:

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog,
and this project adheres to Semantic Versioning.

Unreleased

Changed

• Changed f() and g() into h(b:bool).
• The change is NOT backward compatible!!**
• Please replace .f() -> .h()
• And replace .g() -> .h(b=True)
• Reason: There is a bug in somelib ...
• More: https://bugs.python.org/issueXXXXX
• Release note 2.

1.0.0 - 2017-04-10

Deprecated

• Known issue 1 (1.0.0)
• Known issue 2 (1.0.0)

As we can see, each line is managed as a new note.

import keepachangelog, pprint
pprint.pprint(keepachangelog.to_dict('CHANGELOG.md', show_unreleased=True)['unreleased']['changed'])

['Changed `f()` and `g()` into `h(b:bool)`.',
 'The change is NOT backward compatible!!**',
 'Please replace `.f()` -> `.h()`',
 'And replace `.g()` -> `.h(b=True)`',
 'Reason: There is a bug in `somelib` ...',
 'More: https://bugs.python.org/issueXXXXX',
 'Release note 2.']

[Mulugruntz]

And empty lines (which have their role in Markdown) are striped.

keepachangelog/keepachangelog/_changelog.py

Line 94 in 3538c5b

line = line.strip(" \n")

[Colin-b]

Hello @Mulugruntz,

Thanks for reporting.

This is indeed a known issue. Markdown parsing is basic for now as we treat each line as a separate entry.
However I don't have time to handle this for now, if this is urgent feel free to come up with a pull request :)

Thanks again

[Mulugruntz]

Sure @Colin-b !

I just finished rewriting the whole parser.
For now, it does almost the same as the current one (there were a few weird behaviours/bugs that made little to no sense, I "fixed" them). These quirks can be seen in the test_ files I modified.

There are still some odd behaviours that could be discussed but I chose to not change them, to be as close as possible to the current one.

Anyway, now there is a whole bunch of objects that handle different parts of the changelog. And they can be serialized/deserialized.

I've seen that you did not have a single dependency, so I kept it this way (would have been so much simpler / faster with pydantic though 😞 ).

This is but a first step towards easier maintenance and future development.
Next, I will handle that multiline thing.

You can check the changes so far.
develop...Mulugruntz:new_object_parser

[Mulugruntz]

• Change ordering is now deterministic
  develop...Mulugruntz:new_object_parser#diff-b9882a57033d0b680213990b1ee1d854d5d67b4fb19cd9f2b99a968f67e4dd4eL230-L252
• Release dates are parsed (several formats are supported). When to_dict(), a unique format is chosen.
  develop...Mulugruntz:new_object_parser#diff-a6ac27daa09df0a8c33a7aa07163a5397fcec368bcaebbc939d2cbb0860b902aL89-R89
• When to_raw_dict(), release dates are kept "as-is" (and .lower() is no longer applied)
  develop...Mulugruntz:new_object_parser#diff-a6ac27daa09df0a8c33a7aa07163a5397fcec368bcaebbc939d2cbb0860b902aL266-R271
• When to_raw_dict(), empty lines are preserved
  develop...Mulugruntz:new_object_parser#diff-a6ac27daa09df0a8c33a7aa07163a5397fcec368bcaebbc939d2cbb0860b902aL257-L258
• When semantic version exists, it is produced. Before, if depended on some bugs (it was produced sometimes and not other times, such as when the change was empty or if there was only the link, something like that...). This is now more consistent.
  develop...Mulugruntz:new_object_parser#diff-b9882a57033d0b680213990b1ee1d854d5d67b4fb19cd9f2b99a968f67e4dd4eL127-L128

[Mulugruntz]

Followed-up on #37